In Central America, HIV is transmitted predominantly through unprotected sex. HIV infections are concentrated among specific groups such as sex workers and men who have sex with men. The Garífuna, an ethnic minority group of African Caribbean origin, has higher rates of HIV compared with the general population in Honduras.1 Honduras has the highest concentration of HIV and AIDS cases in the Central American region, with an estimated prevalence of 1.5% in the adult population and 63,000 (range: 35,000-99,000) Hondurans living with HIV at the end of 2005.1 In studies conducted in 2002, men who have sex with men and female sex workers were found to have high HIV prevalence of 13% and 10%, respectively.2,3 Nationally, HIV prevalence among women participating in antenatal care sentinel surveillance studies was reported at 1.4% (n = 3241) in 1998 and 0.5% (n = 5267) in 2004.4
Although there are at least 8 different ethnic minority groups in Honduras,5 information on HIV prevalence is available only for the Garífuna community. A 1999 study among a convenience sample of 310 sexually active participants from urban Garífuna communities revealed an HIV prevalence of 8% and a syphilis seroprevalence of 11%.6 The 1999 study did not collect information on sexual behavior; however, a study conducted in 2000 documented low levels of condom use-less than 5% with regular partners (reported by both men and women) and 25% with casual partners (reported by men only).7
The origin of the Garífuna dates back to the early 1600s when slave ships carrying West Africans shipwrecked near the island of St. Vincent. The survivors mixed with the local Amerindian population, giving birth to the black Caribs or Garinagu, who settled along the Caribbean coast of Central America; Garífuna communities now exist in Belize, Guatemala, Honduras, and Nicaragua. Estimates of the size of the Garífuna population vary from 58,0008 to 250,000.5 Economic changes in the Central American region have led to the growth of Garífuna communities in London and major American cities including New York, New Orleans, and Los Angeles.9-11 The most recent phase in the Garífuna migration is marked by transnational communities with strong ties to Central American communities of origin.12
Our earlier work identified several factors that potentially contribute to a higher HIV prevalence among the Garífuna compared with the general population in Honduras.13 One of these is frequent male migration for seasonal work in the United States or San Pedro Sula (the industrial capital of Honduras) or migration to work in the shipping and fishing industries. Additional factors include high levels of sexual partner violence, alcohol and drug use, inability of women to negotiate safe sex, multiple sexual partners, sexual debut at an early age, reports of informal transactional sex for supplemental income, and infrequent condom use.
To further characterize individual behaviors associated with HIV and other sexually transmitted infections (STI) and establish the current prevalence of HIV and STI in the Garífuna population, we conducted an HIV, STI, and behavioral survey in Honduran Garífuna communities in 2006. This is the first HIV and STI study of a representative sample of the Garífuna population that uses computer-based interviews and assesses HIV and STI prevalence.
Study Population and Enrollment
The Honduran Garífuna population consists of an estimated 200,000 individuals. Garífunas are concentrated in 45 towns and cities in the departments of Atlántida, Colon, Cortes, Gracias a Dios, and Islas de la Bahía. We selected participants from the eight largest Garífuna communities in the three departments with the highest concentrations of Garífunas.5 Both urban and rural areas were represented. We defined a rural community as being located more than one hour by automobile from an urban center.14 We selected a total of five urban communities in the department of Atlántida and a total of three rural communities in the departments of Cortes and Colon (Fig. 1). These communities are populated almost exclusively by Garífunas with the exception of La Ceiba, which is approximately 50% Garífunas.5
We performed multistage cluster sampling in each community. After the completion of community censuses, maps showing individual households were divided into subdistricts according to neighborhood delineations, and the number of households per subdistrict was noted. Subdistrict populations ranged from 200 to 790 individuals. According to the study census, 55% of the population in the selected communities was female and 45% was male. Subdistricts served as the primary sampling units and were randomly selected proportional to size based on number of households. Individual households within each primary sampling unit were then selected and visited by study interviewers. If the residents of a selected household were not of Garífuna ethnicity, the neighboring household was chosen. The male adult (≥18 years of age) residing in the household who most recently celebrated his birthday was asked to participate in the survey. This procedure was also used to select an adult female from the same household. If the household contained either 1 eligible female or 1 eligible male but not both, the eligible person was invited to participate. If the household contained neither an eligible female nor an eligible male, the neighboring household was selected. If either of the selected adults was not present, the interviewer asked when she/he could return to speak to the selected person. If the selected person was not available at the second visit, the neighboring household was chosen and the procedure was repeated. If the selected person refused to participate, the household was considered a refusal household and no further attempt was made to recruit an eligible adult from that household. Appointments were made with all eligible participants to conduct interviews at a community health center or nongovernmental organization office, that provided adequate space and privacy to survey administration, counseling, and sample collection.
We planned for a sample size of 800 Garífuna adults: 400 women and 400 men, with half of the participants from urban sites and half from rural sites. The total sample size was designed to allow a comparison between the current survey and a follow-up survey to be conducted after four years, to detect a change of at least 15% on a behavioral indicator. The indicator chosen was condom use during last sexual intercourse with any type of partner, assuming, conservatively, a baseline level of 50%. The accuracy level was 5%, and the power to detect a difference was 80%. The estimated design effect was 1.25. A sample size of 400 participants would be sufficient to estimate an HIV prevalence of 5.0% with a 95% confidence interval (CI) of 3.0% to 6.7%.
Written informed consent was obtained from all participants. Each interview included two components: (1) completion of a behavioral questionnaire via audio computer-assisted interviews (ACASI) and (2) collection of specimens for STI and HIV tests. Participants could refuse participation in either component. Interviewers were peer outreach workers selected from HIV prevention organizations and received training on study procedures, consent administration, use of ACASI, interviewing techniques, and confidentiality. They were supervised by site coordinators and regional coordinators. Using ACASI, participants read the survey questions displayed in Spanish on a computer screen while listening to audio recordings of the questions using headphones. They entered their answers using the computer keyboard. The study questionnaires were based on standardized survey instruments15 and elicited information on the following demographics: sexual behavior in the past year, including number and type of sexual partners; sexual practices and condom use; knowledge of condoms; knowledge about STI and HIV; symptoms and health-seeking behaviors; and use of alcohol and drugs, including injected drugs.
After completion of the consent process and of the computer interview, trained counselors provided HIV and STI pretest counseling and collected biological specimens. A nurse or doctor conducted a physical exam to evaluate signs and symptoms of STI. Female participants received a pelvic exam during which vaginal swabs were collected. Male participants provided a urine specimen. Blood specimens were collected by a trained phlebotomist. Urine (for men) and vaginal samples were screened with polymerase chain reaction for bacterial STI, including Chlamydia trachomatis, Neisseria gonorrhoea, Trichomonas vaginalis, and Mycoplasma genitalium. Blood samples were tested for HIV, syphilis, and herpes simplex type 2 (HSV-2). HIV testing was performed on site with two rapid HIV tests (Determine HIV-1/2 and OraQuick Rapid HIV-1 antibody test). Discordant samples were sent to a reference laboratory for enzyme-linked immunosorbent assay testing. Syhilis seropositivity was estimated based on reactive rapid plasma reagin and positive Treponema particle agglutination assay results. Active syphilis was defined as a rapid plasma regain titer ≥1:8 and a reactive Treponema particle agglutination assay result. HSV-2 serology was performed with the HerpeSelect (Focus Technologies, Cypress, CA). Participants received posttest counseling and HIV test results 30 minutes after testing. STI test results were provided four weeks later through the participating government health centers. Free treatment was provided to participants with STI symptoms or a laboratory-diagnosed STI.
HIV enzyme-linked immunosorbent assay for discordant results, syphilis, and HSV-2 serologies were done at the National Reference Laboratory in Tegucigalpa, Honduras. Quality control and polymerase chain reaction testing was done at the Centers for Disease Control and Prevention Laboratory in Atlanta, Georgia.
The study protocol was approved by the Ethics Review Committee of the Autonomous National Honduras University and was reviewed by the Global AIDS Program Associate Director of Science who deferred approval to the Honduras Ethics Committee.
Data analysis was performed using Stata software version 9.0 (STATA, College Station, TX). Sampling weights, applied to all analysis reported unless otherwise indicated, were calculated as inverse sampling fractions by sex for each community. Sampling fractions across communities ranged from 0.05 to 0.35 among women and from 0.04 to 0.38 among men. The use of sampling weights allows respondents to be considered as representative of the total number of male and female residents in the community sampled. However, because we did not collect information that would allow respondents to be linked with households, we cannot account for differences in probabilities of selection due to differences in the proportion of persons in each household sampled. Consequently, the true level of sampling error is unknown. We researched sampling errors for studies in Central America with similar study designs. We assumed a similar level of sampling error as that estimated in the same departments (Atlántida, Cortés, and Colón) in the 2005 National Demographic and Health Survey. We report confidence intervals (CI) that are expanded to reflect a design effect of 1.8. This is the average of the maximum design effect across variables for the 3 departments in the 2005 Demographic and Health Survey.16 The criteria for statistical significance was set at a P value of ≤0.009, which corresponds to a design effect of 1.8 for a 2-tailed t test [2 (1−Zsqrt(1.8)× 1.96)].
We estimated prevalence of infection for HIV and positivity for other STI and proportions for demographics and risk behaviors. Urban-rural and sex differences in sample proportions were assessed using a Pearson χ2 test for dichotomous variables. Unweighted K sample tests were used to test for differences in continuous variables (eg, age of first sex). Unweighted T tests of mean differences yielded equivalent results.
Odds ratios from logistic models were used to assess bivariate and multivariate associations of HIV and HSV-2 with demographics and risk behaviors among sexually experienced participants (ie, those reporting any previous sexual relation). Separate multivariate logistic models predicting HIV and HSV-2 infection were developed. Multivariate models initially included all variables with a bivariate association at the 5% level and fixed effects for community of residence. Second-order interactions between covariates and fixed effects were retained if they improved global goodness of fit of the weighted model at the 5% level. Square root, natural logarithm, and second-order quadratic transformations of centered continuous covariates (age and age at first sex) were evaluated similarly.
Of 880 people approached, 824 (94%) agreed to participate in the study. Seven individuals were excluded from analysis because of inadequate labeling of questionnaires or laboratory samples. Of the 817 participants remaining, 15 refused to be tested for HIV and 25 did not provide specimens for STI testing. Three hundred ninety-nine participants (49%) were women and 418 (51%) were men; 399 (49%) were from rural sites and 418 (51%) were from urban areas. All participants received their HIV test results and 82% returned for their STI results. The following estimates of demographics and behaviors are weighted so as to be representative of the population.
The estimated population median age was 30 years [interquartile range (IQR): 22-44 years]. Thirty percent had less than a primary education, 35% had completed a primary education, and 34% had completed more than a primary education. There were no significant differences between men and women by education. Twenty-eight percent of men and 37% of women had an income below 500 Lempiras (US $27) per month (P = 0.04). Forty-eight percent of men were employed compared with only 26% of women (P < 0.001), and 17% of men and 26% of women received remittances from abroad as their main income (P = 0.01). Sixty percent of men and 33% of women consumed alcohol in the past month (P < 0.001); 20% of men and 5% of women used illegal drugs in the past month (P < 0.001) (Table 1).
The estimated population median age of first sex was 15 years (IQR: 13-17 years) for men and 17 years (IQR: 15-19 years) for women (P < 0.001). Overall, 78% of men and women had sexual intercourse in the past year. Fifteen percent of men and 18% of women aged 18 and older reported no sexual experience during their lifetime, precluding them from responding to sexual behavior questions. Five percent of men and 2% of women had paid for sex in the past 12 months, and 12% of men and 7% of women had received money for sex in the past 12 months. Five percent of men and 3% of women had sex with a same-sex partner in the past year (P = 0.4). Both men (8%) and women (12%) had been coerced into sex in the past year (P = 0.2). Condom use during last sex with a stable partner was 12% among men and 10% among women, whereas condom use during last sex with a casual partner was 44% and 36% for men and women, respectively. Multiple partners (>1 in the past year) were reported by 57% of men and 44% of women (P = 0.07). Only 9% of men were circumcised (Table 1).
Forty-one percent correctly identified three or more routes of HIV transmission and 23% correctly identified all 10 transmission routes. The questions that were most commonly answered incorrectly included the following: (1) can HIV be transmitted from an HIV-infected pregnant mother to her baby? (61% incorrect responses); (2) can HIV be transmitted by receiving a blood transfusion with contaminated blood products? (45% incorrect responses); and, (3) can HIV be transmitted by reusing a needle that has been used by someone HIV infected? (58% incorrect responses). There were no significant differences in the number of correct responses between men and women or between respondents from urban vs. rural areas.
Estimated population HIV prevalence was 5%; 4% in urban areas and 5% in rural areas (P = 0.9), and 4% among men and 5% among women (P = 0.4) (Table 2). Estimated population HSV-2 seropositivity was 51% overall and was significantly higher among women (60%) than men (41%) (P < 0.001) and in rural (61%) vs. urban areas (48%) (P < 0.001). Overall, syphilis seropositivity was 2% and active syphilis was 0.5%. Positivity for other STI were 1% gonorrhea, 7% chlamydia, 7% for mycoplasma, and 11% for trichomoniasis. Trichomoniasis was higher among women (16%) than men (4%) (P < 0.001).
There were differences in STI positivity between men and women in urban and rural areas. In urban areas, there were differences between men and women on HSV-2 seropositivity (56% women vs. 38% men; P = 0.002), syphilis seropositivity (1% women vs. 5% men; P = 0.04), gonorrhea (3% women vs. 1% men; P = 0.0010), and trichomoniasis (15% women, vs. 4% men; P = 0.001). In rural areas, we found significant differences for HSV-2 (70% women vs. 50% men; P < 0.001) and trichomoniasis (21% women vs. 5% men; P < 0.0001).
Among respondents who reported no sexual experience, nine tested positive for HIV (three men and six women). Respondents who reported never having had sex were older (P < 0.001), less likely to be employed (P = 0.004), and less likely to use illicit drugs (P = .004) than sexually experienced respondents.
Variables associated with HIV infection among sexually experienced individuals in the bivariate analysis at the 5% level were: income less than 500 lempiras (P = 0.02), consistent condom use with stable partner (P = 0.04), and self-reported STI in past 12 months (P = 0.01) (Table 3). At the stricter 0.9% level, significant factors were: marijuana use in the past 12 months (P = 0.007) and HSV-2 infection (P = 0.006). In the final multivariate model, an interaction between low income and urban residence was significant (P = 0.005), highlighting higher levels of HIV (8%) among urban, poor Garífunas.
Variables associated with HSV-2 seropositivity among sexually experienced individuals in the bivariate analysis at the 5% level were consistent condom use with stable partner (P = 0.03) and T. vaginalis infection (P = 0.03) (Table 4). Age (P = 0.001), being female (P < 0.001), urban residence (P = 0.002), age at first sex older than 15 years (P = 0.009), syphilis seropositivity (P < 0.001), and HIV infection (P = 0.006) met the stricter 0.9% criterion. Community effects were also identified; Limón (P = 0.02), Sambo Creek (P = 0.005), Santa Rosa de Aguán (P = 0.02), and Tornabe (P < 0.001) had higher HSV-2 prevalences than other study sites. In the multivariate analysis, variables that remained significantly associated with HSV-2 seropositivity included being female (P < 0.001), older age (P < 0.001), and syphilis seropositivity (P = 0.007). Results from the multivariate model indicate that the odds of HSV-2 infection increase 10% for an increase of 1 year of age, at the mean age (37 years among sexually experienced individuals in the sample). The association with HIV was of borderline significance (P = 0.017) but did not meet the stricter 0.9% criteria. Goodness of fit of the HIV and HSV-2 models was supported at levels of P = 0.893 and P = 0.442, respectively.
The Garífuna population in Honduras has a substantial burden of HIV and other STI. This study identified several factors associated with these infections and thus provides direction for future program priorities. This is the first HIV and STI study of a representative sample of the Garífuna population to have used ACASI to reduce reporting bias. The results show that it was common for both men and women to have multiple partners and that condom use with stable and casual partners was alarmingly low. After adjusting for other variables and confounders, we found that poor people in urban areas had significantly higher rates of HIV infection.
Although paying for sex was not common in this population, 6% had received money for sex. This behavior was more common among men than women. Receiving money for sex was not a significant predictor of HIV infection in this study but has the potential to facilitate transmission. In a recent qualitative study conducted among the Garífuna in preparation for this survey, exchanging sex for money was frequently reported. The authors13 suggest that economic deprivation influences decisions to purchase condoms, engage in commercial sex work, or enter into sexual relationships to obtain other resources (such as “in style” clothing or electronic appliances). Many Garífuna relationships reportedly involve women who seek men to provide resources for themselves and their children.
A 1999 convenience sample of 310 sexually active Garífunas from 4 urban communities found an HIV prevalence of 8% (95% CI: 6 to 12), syphilis seroprevalence of 11% (95% CI: 8 to 15), and active syphilis prevalence of 1% (95% CI: 0.5 to 4), all of which are higher than the prevalences reported in this study.6 HIV prevalence was also consistently lower at each site in the current study than in the 1999 study, which included the same study sites. Data from a behavioral survey conducted in 2000 suggest that there may also have been changes in condom use. The 2000 study used a multistage stratified random sample from five Garífuna communities, four of which were also included in the current study; weighted point estimates and CI were not presented in the 2000 report.7 However, the average age of sexual debut was similar in both studies (15 years of age for men and 17 years for women). The proportion reporting more that 1 sexual partner in the preceding year was 28% of men and 3% of women in 2000 compared with 36% and 29%, respectively, in this study. Consistent condom use with a stable partner was 4% in men and 5% in women compared with 12% and 10%, respectively, in the current study. Consistent condom use with a casual partner was reported by 25% of men in 2000 and 42% of men in the current study; this variable was not reported for women in 2000. In summary, our survey found lower levels of syphilis and HIV prevalence, higher levels of condom use, and no difference with respect to fidelity and abstinence compared with the earlier studies. However, substantial differences in methodology between the surveys limit comparability. Future surveillance studies using probabilistic samples are needed to confirm this trend.
We found that age, being female, lower education, and syphilis were all associated with HSV-2 in multivariate analysis. After adjusting for the study design effect, the association of HSV-2 and HIV infection was of borderline significance. Previous studies have described similar associations between herpes and these variables,17-20 with the exception of syphilis. Because this is a cross-sectional survey, the association may be partially explained by shared risk factors and transmission routes. Because HSV-2 has been linked to HIV acquisition, studies assessing the cofactor effect of syphilis on HIV acquisition should control for the presence of HSV-2. HSV-2 prevalence in this population was distressingly high-comparable to what has been reported for the general population in Africa21 and higher than other reports available for the general population in Latin America. HSV-2 prevalence among women from the general population has been reported at 23% in Brazil and 30% in Mexico.22,23 There are no reports of HSV-2 prevalence among men in the Latin American region. HSV-2, even in the absence of active lesions, is a recognized risk factor for HIV acquisition and transmission and may be contributing to the higher HIV prevalence among the Garífuna.19
This study has several limitations. Relatively few individuals were HIV positive, which limits the power to analyze risk factors for infection. Odds ratios from variables identified in the HIV risk factor analysis have wide CI, so the magnitude of the estimates should be interpreted with care. Several individuals with HIV-positive and STI-positive test results reported that they had never had sexual intercourse and were therefore excluded from the risk factor analysis; this suggests either misclassification by participants due to social desirability bias or less than 100% specificity of the test results. Positive predictive values of a test decrease when populations with low prevalence are tested. Misclassification by study participants has been reported frequently in population-based surveys and other behavioral surveys among specific groups and is a common limitation of research that relies on self-reported data.24,25 The high enrollment shows the feasibility and acceptability of the ACASI method for collecting data from this population. However, it must be noted that the use of ACASI may lead to an increased, but not necessarily complete, reporting of stigmatized behaviors. The misclassification of participants by sexual activity may have introduced a negative bias leading to underestimation of high-risk behaviors and a decreased ability to detect significant associations with HIV and HSV-2.
Interventions to increase patient and physician awareness of herpes symptoms and the availability of herpes treatment should be considered for this population. Other interventions for HIV and STI prevention should include condom promotion programs that have a culturally appropriate approach and are delivered in both Garífuna and Spanish languages. Although almost half of the study population reported having ever been tested for HIV, programs to increase coverage of and access to testing are warranted. In particular, concerted efforts to raise awareness of HIV and its modes of transmission are needed to actively dispel misconceptions regarding HIV infection and spread.
We would like to express our gratitude to the Department of Health of Honduras for leading this study and providing permission to use government health facilities. We would especially like to thank the staff at the participating clinics. We are deeply grateful to the study participants who made this survey possible. We have furthermore to thank Ron Ballard, Celine Taboy, and Lisa Steele for processing the specimens at the Centers for Disease Control and Prevention STD Laboratory and providing technical advice and Meade Morgan for his statistical advice. Many thanks to the United States Agency for International Development office in Honduras and the Centers for Disease Control and Prevention for funding the study. Thanks to Kelly Stewart from United States Agency for International Development for her advice and support. We would finally like to thank Mirim Sabin and George Luber for conducting the qualitative study that helped the development of this survey.
1. UNAIDS, ed. AIDS Epidemic Update: December 2006
. Geneva, Switzerland: Joint United Nations Program on AIDS (UNAIDS); 2006: 1-23.
2. Padilla I, Soto R. Estudio multicéntrico centroamericano de prevalencia de VIH/ITS y comportamientos en hombres que tienen sexo con otros hombres en Honduras
. Tegucigalpa, Honduras: AIDS Action Project for Central America-PASCA; 2003.
3. Paredes M, Soto R. Estudio multicéntrico centroamericano de prevalencia de VIH/ITS y comportamientos en mujeres trabajadoras comerciales del sexo en Honduras
. Tegucigalpa, Honduras: AIDS Action Project for Central America-PASCA; 2003.
4. Lara B, Gupta S, Aragon M, Monterroso E, Paredes M. El perfil epidemiológico de VIH/SIDA en Honduras
. Tegucigalpa, Honduras: Departamento de ITS/VIH/SIDA, Secretaria de Salud de Honduras; 2006.
5. Lara Pinto G, ed. Perfil de de los Pueblos Indígenas y Negros de Honduras
, Unidad Regional de Asistencia Tecnica (RUTA). Tegucigalpa, Honduras: Banco Mundial; 2002:153.
6. Sierra MA, Paredes M, Pinel R, et al. Estudio seroepidemiologico de Sífilis, Hépatitis B y VIH en poblacion Garífuna de El Triunfo de la Cruz, Bajamar, Sambo Creek y Corozal
. Tegucigalpa, Honduras: Secretaria de Salud Publica, Honduras; 1999.
7. García JM, Salvarria N, Valentin D, et al. Aspectos sociodemográficos, conocimientos, creencias, actitudes y prácticas relacionadas con la transmisión de ITS/VIH en las comunidades garífunas de: Travesía, Bajamar, El Triunfo de la Cruz, Santa Fe y Limón
. Tegucigalpa, Honduras: Asociación El Buen Pastor, Organización Panamericana de la Salud/ONUSIDA, Departamento de ETS/SIDA/TB, Secretaría de Salud de Honduras; 2000.
8. INE. Population projections 2001-2015
. Tegucigalpa, Honduras: National Statistics Institute; 2001.
9. Gonzalez NLS. Sojourners of the caribbean: Ethnogenesis and ethnohistory of the Garífuna
. Urbana, IL: University of Illinois Press; 1988.
10. Rivas RD. Pueblos indigenas y garifuna de Honduras
. Tegucigalpa, Honduras: Editorial Guaymuras; 2000.
11. Suazo ES. Los deportados de san vicente
. Tegucigalpa, Honduras: Editorial Guaymuras; 1997.
12. Stansbury JP, Sierra M. Risks, stigma and Honduran Garifuna conceptions of HIV/AIDS. Soc Sci Med
13. Sabin M, Luber G, Sabin K, et al. Rapid ethnographic assessment of HIV/AIDS among Garífuna communities in Honduras: Informing HIV Surveillance among Garífuna women
. J Hum Behav Soc Environ
14. Villalvazo Peña P, Corona Medina J, García Mora S. Urbano-rural, constante búsqueda de fronteras conceptuales
. Revista de Información y Análisis. Mexico DF, Mexico: Instituto Nacional de Estadistica, Geografia Informática (INEGI), 2002.
15. FHI. Behavioral surveillance surveys (BSS) guidelines for repeated behavioral surveys in populations at risk of HIV
. Arlington, VA: Family Health International; 2006.
16. INE. Encuesta Demografica de Salud en Hombres, ENDESA 2007
. Tegucigalpa, Honduras: Instituto Nacional de Estadistica; 2007.
17. Weiss HA, Buve A, Robinson NJ, et al. The epidemiology of HSV-2 infection and its association with HIV infection in four urban African populations. AIDS
. 2001;15 (Suppl 4):S97-S108.
18. Paz-Bailey G, Ramaswamy M, Hawkes S, Geretti AM. Herpes simplex virus type 2: Epidemiology and management options in the developing countries. Sex Transm Infect
19. Freeman EE, et al. Herpes simplex virus 2 infection increases HIV acquisition in men and women: systematic review and meta-analysis of longitudinal studies. AIDS
20. Wald A, Link K. Risk of human immunodeficiency virus infection in herpes simplex virus type 2-seropositive persons: a meta-analysis. J Infect Dis
21. Weiss H. Epidemiology of herpes simplex virus type 2 infection in the developing world. Herpes. 2004;11(Suppl 1):24A-35A.
22. Carvalho M, de Carvalho S, Pannuti CS, et al. Prevalence of herpes simplex type 2 antibodies and a clinical history of herpes in three different populations in Campinas City, Brazil. Int J Infect Dis
23. Lazcano-Ponce E, Smith JS, Munoz N, et al. High prevalence of antibodies to herpes simplex virus type 2 among middle-aged women in Mexico City, Mexico: a population-based study. Sex Transm Dis
24. Catania JA, Gibson DR, Chitwan DO, Coates TJ. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull
25. IMPACT/FHI/UNAIDS. Chapter 5: Do People Tell the Truth About Their Sexual and Drug-Taking Behaviour? In Meeting the Behavioral Data Collection Needs of National HIV/AIDS and STD Programme
. Washington, DC: IMPACT/FHI/UNAIDS; 1998.